Cognotik | PoliticalOptimizationTask

⚙️ ExecutionConfig.json

{
  "initial_text": "Universal basic income is...",
  "optimization_goal": "maximize consensus",
  "perspectives": [
    "progressive", "conservative", 
    "libertarian", "centrist"
  ],
  "evaluation_criteria": ["clarity", "persuasiveness"],
  "population_size": 8,
  "consensus_mode": "maximize",
  "num_generations": 5
}

→

👁️ Consensus Analysis UI

Consensus Score 74.2 (Unifying)

Average Quality 88/100

COMMON GROUND

Economic security in automation era
Reducing administrative overhead

POINTS OF CONTENTION

Funding via wealth tax vs. VAT
Impact on labor participation

Test Workspace Browser

Explore actual artifacts, logs, and generated variants from the PoliticalOptimizationTask workspace.

Execution Configuration

Field	Type	Description
`initial_text`*	String	The initial text to analyze or optimize.
`optimization_goal`*	String	The goal (e.g., 'maximize consensus', 'identify wedge issues').
`perspectives`	List<String>	Perspectives to evaluate from. Default: progressive, conservative, libertarian, centrist.
`evaluation_criteria`	List<String>	Criteria like 'clarity', 'persuasiveness', 'factual_accuracy'.
`consensus_mode`	String	'maximize' (unify), 'minimize' (divide), or 'explore' (both).
`population_size`	Int	Number of variants per generation. Default: 8.
`num_generations`	Int	Number of evolutionary cycles. Default: 5.
`consensus_weight`	Double	Weight (0.0-1.0) of consensus in fitness calculation. Default: 0.6.
`mutation_strategies`	List<String>	Strategies: 'rephrase', 'emphasize', 'soften', 'reframe'.
`enable_crossover`	Boolean	Whether to combine high-performing variants. Default: true.

Task Lifecycle

Initialization: Validates configuration (min 2 perspectives, valid consensus mode).
Multi-Perspective Evaluation: The LLM assumes each political persona to score text across criteria.
Consensus Scoring: Calculates signed variance where low variance = high consensus (positive) and high variance = divisive (negative).
Evolutionary Loop: Performs selection, mutation (rephrase/polarize/bridge), and crossover across multiple generations.
Reporting: Generates detailed analysis of common ground, points of contention, and strategy effectiveness trends.

Embedded Execution (Headless)

Use the UnifiedHarness to run this task in CI/CD or automated scripts without a UI.

import com.simiacryptus.cognotik.apps.general.UnifiedHarness
import com.simiacryptus.cognotik.plan.tools.social.PoliticalOptimizationTask
import com.simiacryptus.cognotik.plan.tools.social.PoliticalOptimizationTask.PoliticalOptimization

val harness = UnifiedHarness(serverless = true, openBrowser = false)
harness.start()

val config = PoliticalOptimizationTask.PoliticalOptimizationTaskExecutionConfigData(
    initial_text = "Universal basic income should be funded by a land value tax.",
    optimization_goal = "maximize consensus",
    perspectives = listOf("progressive", "conservative", "libertarian"),
    consensus_mode = "maximize",
    num_generations = 3
)

harness.runTask(
    taskType = PoliticalOptimization,
    executionConfig = config,
    workspace = File("./political-analysis")
)

Gradle Dependency

dependencies {
    implementation("com.cognotik:webapp:2.0.39")
}

CLI / GitHub Action Example

Run as a standalone tool to generate a consensus report.

# Run via Cognotik CLI
java -jar cognotik-cli.jar \
  --task PoliticalOptimization \
  --initial_text "Carbon taxes are the most efficient way to reach net zero." \
  --optimization_goal "identify wedge issues" \
  --consensus_mode "minimize"

Prompt Segment

The following logic is injected into the LLM context:

PoliticalOptimization - Optimize text using multi-perspective political consensus analysis
  ** Specify the initial text to analyze/optimize.
  ** Define political perspectives to evaluate from (progressive, conservative, libertarian, centrist, etc.).
  ** Set optimization goal (maximize consensus, minimize divisiveness, or explore both).
  ** Configure evaluation criteria (clarity, persuasiveness, factual accuracy, emotional appeal, etc.).
  ** Choose consensus mode: maximize (unify), minimize (divide), or explore (both).
  ** The task will:
     - Evaluate text from each political perspective independently.
     - Calculate consensus score (positive = unifying, negative = divisive).
     - Identify common ground and points of contention.
     - Generate variants optimized for consensus or division.
     - Track evolution of agreement/disagreement.